Occasionally, you will get a funny looking image which looks as if one part has slipped relative to the other, or has bars across it or some equally obvious fault. The most likely cause is a fault occurring during the transfer of the image from detector to disc. In order to check this, look a few lines back on the ADAM screen for a line to the effect:
`Time-out during DMA transfer'
If you do find this message then the image is hopelessly corrupted and cannot be recovered. The observation is unfortunately lost; you may as well PURGE the file at once, and use the NEXTRUN command to reset the run number if you do not want an hiatus in the numbering of your observations. After a failed DMA transfer the next observation usually transfers without difficulty. You can do a CCDINIT, but there is no clear evidence that this makes subsequent transfers any more reliable.
However, if you do NOT find an error message about DMA transfer problems, then your image has probably been written on to an irrecoverably corrupt sector of the EGLE disc. In this case do NOT purge or delete the file! Instead, leave the file as it is, but don't copy it to tape, because the image is corrupted forever and you cannot do anything about it, except repeat the observation. Further, leave a note to the software group to the effect that file Rn on account m is written on a corrupt sector of the EGLE. They will then rename the file to a hidden account, leaving the data on the disc, which effectively prevents ADAM from trying to use that part of the disc again. Therefore, whatever you do, DO NOT DELETE the file, as the software group won't be able to find the corrupted area again until some unfortunate following observer, or indeed you yourself, succeeds in putting a (necessarily corrupted) image on it. By following this policy quite a number of bad sectors on the disc are now protected from being used by ADAM, and faults of this nature are now rarely encountered.
The worst variation of the aforementioned problem is that your GLANCE files are written on a bad sector: doing a new GLANCE simply will write on the same bad sector. The only way out is to KEEP the glance file with the faulty image, which is then renamed to Rn, and inform the software staff as before about the corrupted file. Subsequent GLANCEs will be directed to a different part of the disc.